A method for the detection of micromotion of a metal-backed hemispherical acetabular cup is presented and tested. Unlike in conventional roentgen stereophotogrammetric analysis, the cup does not have to be marked with tantalum markers; the micromotion is calculated from the contours of the hemispherical part and the base circle of the cup. In this way, two rotations (tilt and anteversion) and the translations along the three cardinal axes are obtained. In a phantom study, the maximum error in the position of the cup's centre was 0.04 mm. The mean error in the orientation of the cup was 0.41 degree, with a 95% confidence interval of 0.28-0.54 degree. The in vivo accuracy was tested by repeated measurement of 21 radiographs from seven patients. The upper bound of the 95% tolerance interval for the translations along the transversal, longitudinal, and sagittal axes was 0.09, 0.07, and 0.34 mm, respectively: for the rotation, this upper bound was 0.39 degree. These results show that the new method, in which the position and orientation of metal-backed hemispherical cup is calculated from its projected contours, is a simple and accurate alternative to attaching markers to the cup.